and Management 403 (2017) 145–151

Contents lists available at ScienceDirect

Forest Ecology and Management

journal homepage: www.elsevier.com/locate/foreco

Long-term dynamics and characteristics of snags created for wildlife habitat MARK ⁎ Amy M. Barrya, Joan C. Hagarb, James W. Riversa, a Department of Forest Ecosystems and Society, Oregon State University, Corvallis, OR 97331, USA b US Geological Survey, Forest and Rangeland Ecosystem Science Center, Corvallis, OR 97331, USA

ARTICLE INFO ABSTRACT

Keywords: Snags provide essential habitat for numerous organisms and are therefore critical to the long-term maintenance Cavity-nesting birds of forest biodiversity. Resource managers often use snag creation to mitigate the purposeful removal of snags at Created snags the time of harvest, but information regarding how created snags change over long timescales (> 20 y) is absent Oregon Coast Range from the literature. In this study, we evaluated the extent to which characteristics of large (> 30 cm diameter at Pseudotsuga menziesii breast height [DBH]) Douglas-fir(Pseudotsuga menziesii) snags created by topping had changed after 25–27 y. We also tested whether different harvest treatments and snag configurations influenced present-day snag char- Snag persistence acteristics. Of 690 snags created in 1989–1991, 91% remained standing during contemporary surveys and 65% remained unbroken along the bole. Although most snags were standing, we detected increased bark loss and breaking along the bole relative to prior surveys conducted on the same pool of snags. Although snag char- acteristics were not strongly influenced by snag configuration, we found that snags in one harvest treatment (group selection) experienced less bark loss and had lower evidence of use by cavity-nesting birds (as measured by total cavity cover) relative to snags created with clearcut and two-story harvest treatments. Our results indicate that Douglas-fir snags created by topping can remain standing for long time-periods (≥25 y) in man- aged , and that the influence of harvest treatment on decay patterns and subsequent use by wildlife is an important consideration when intentionally creating snags for wildlife habitat.

1. Introduction Waldien et al., 2006; Kluber, 2007; Kilgo and Vukovich, 2014). Despite their ecological significance, snags are often removed Standing dead , or snags, are important ecological structures during timber harvest because of their commercial value and to comply that can be formed naturally through the actions of fire, wind, , with safety regulations (McComb et al., 1993; Kroll et al., 2012). and fungi (Morrison and Raphael, 1993; Rose et al., 2001). Snags play a Therefore, decades of intensive practices, such as crucial role in forest health by storing terrestrial carbon and con- and fire suppression, have resulted in a reduction in the tributing to soil development as snags decompose (Harmon et al., 1986; number of snags within managed forest landscapes (Cline et al., 1980; Rose et al., 2001; Angers et al., 2012). In addition, they enhance bio- Swanson and Franklin, 1992; Wilhere, 2003). A reduction in snags can diversity by providing nesting and foraging habitat for many species of reduce native biodiversity by degrading or eliminating habitat (Hane insects, amphibians, birds, and mammals during their transition from a et al., 2012). Therefore, intentional creation of snags, which is typically living to a snag (Thomas, 1979; Harmon et al., 1986; Newton, undertaken at the time of harvest, has been implemented to mitigate 1994; Rose et al., 2001; Seibold et al., 2016). For example, decay in loss of wildlife habitat in managed forests (Bull and Partridge, 1986). living trees and snags creates pockets of rot that provide critical nesting The amount of time that created snags remain standing and the rate and roosting habitat for species that require cavities (e.g., - at which snags decay are critical for determining their useful lifespan as peckers) and areas under loose bark (e.g., bats, Brown Creeper [Certhia wildlife habitat, and snag lifespan can be influenced by the density of americana]; Harmon et al., 1986; Franklin et al., 1987; Chambers et al., residual live trees and the configuration of snags following harvest. 2002; Bunnell, 2013; Geleynse et al., 2016). As snag decay progresses, Direct exposure to environmental factors, such as solar radiation, pre- parts of the snag begin to break and fall to the ground and become cipitation, and wind can cause rapid changes in moisture content of downed wood, providing additional shelter and feeding substrates for wood within snags, in turn influencing bark and wood retention. many species of insects, amphibians, and mammals (Franklin et al., Therefore, greater residual live tree density around snags may decrease 1987; Duane, 2001; Manning and Edge, 2004; Rose et al., 2001; breaking rates and increase bark and wood retention by protecting

⁎ Corresponding author at: Department of Forest Ecosystems and Society, 321 Richardson Hall, Oregon State University, Corvallis, OR 97331, USA. E-mail address: [email protected] (J.W. Rivers). http://dx.doi.org/10.1016/j.foreco.2017.07.049 Received 27 May 2017; Received in revised form 25 July 2017; Accepted 27 July 2017 0378-1127/ © 2017 Elsevier B.V. All rights reserved. A.M. Barry et al. and Management 403 (2017) 145–151 snags from environmental factors (Harmon et al., 1986). Whether snags 2. Methods are retained in clusters or scattered throughout a harvest unit may also affect their decay rate; clustered snags may promote greater co- 2.1. Site description lonization and subsequent mineralization of wood and colonization by fungi, and other invertebrates (Chamberlin, 1918; Angers Our study sites were located within Oregon State University's et al., 2012). The presence of insects can also attract additional species McDonald-Dunn Research Forest (123°15′W, 44°35′N) near Corvallis, such as woodpeckers, which further contribute to snag decay through Oregon on the lower east slope of the Coast Range. The original study fragmentation of wood during their foraging and excavating activities design comprised 30 stands (5–18 ha each) dominated by Douglas-fir (Harmon et al., 1986) and can facilitate the establishment of fungi (Pseudotsuga menziesii) that had regenerated naturally after harvest. (Jusino et al., 2016). Although wildlife use of snags tends to increase These stands included two understory plant association types: hazel with snag age (Schreiber and DeCalesta, 1992; Chambers et al., 1997; (Corylus cornuta var. californica)/brome (Bromus vulgaris) and vine Hallett et al., 2001; Walter and Maguire, 2005; Arnett et al., 2010), maple (Acer circinatum)/salal (Gaultheria shallon; Franklin and Dyrness information is lacking about how long snags persist, and how harvest 1973). Dominant trees on stands were 45–150 y old at the time of type and snag configuration may influence snag characteristics that are harvest treatments were applied and snag creation took place (see important for wildlife. Nevertheless, this information is critical for re- below), and stands were similar in plant species composition among source managers who are charged with creating snags as a form of treatments prior to harvest (Chambers, 1996). Mean density of live habitat mitigation for snag-associated wildlife species within managed conifers was approximately 540 trees/ha, live hardwood tree density forest landscapes (Kroll et al., 2012). averaged 165 trees/ha, and natural snag (i.e., snags formed naturally In this study, our goal was to quantify characteristics of snags that without tree topping) density averaged < 1.9 snags/ha (Chambers, were intentionally created in 1989–1991 as part of a long-term silvi- 1996). At the time of our research (2015–2016), natural snag densities cultural experiment at Oregon State University (i.e., College of varied relative to harvest treatment: group selection treatment: Integrated Research Project; Maguire and Chambers, 2005). Snags were 10.6 snags/ha (SE ± 1.9, n = 16 stands), two-story treatment: created at the time of harvest in one of three harvest treatments (group 6.4snags/ha (SE ± 2.2, n = 7 stands), clearcut treatment: 13.3 snags/ selection: uneven-aged with 33% tree volume removed in patches, two- ha (SE ± 5.1, n = 3 stands; Barry, 2017). story: even-aged with 75% tree volume removed uniformly, clearcut: even-aged with all tree volume removed except for 1.3 live trees/ha) and one of two spatial configurations (clustered, scattered), both of 2.2. Study design which were applied at the stand level (Chambers et al., 1997). Our objectives were to (1) test whether experimental harvest treatment and Our study consisted of a randomized, complete block design with snag configuration influenced persistence of created snags, (2) quantify three study blocks, with each block harvested in a separate year (i.e., how current-day snag characteristics related to decay were influenced Lewisburg block in 1989, Peavy block in 1990, Dunn block in 1991) by harvest treatment and snag configuration, and (3) incorporate his- and planted the spring following harvest. Each block contained n = 10 toric data from the same pool of snags to document changes in char- stands in which snags were created at the time of harvest. Individual acteristics and use by cavity-nesting birds across a ≥25-y period. stands within each block were assigned randomly to one of three har- Experimental harvest treatments in our study differed in the amount vest treatments and one of two snag configurations. The harvest treat- of basal area removed, so we hypothesized that greater harvest in- ments and snag configurations were intended to mimic variations in tensity would result in greater levels of decay, as tree density is thought natural disturbance patterns and test the effects of operational alter- to influence decay rate (Garber et al., 2005; Harmon et al., 1986; natives to traditional clearcutting on a range of ecological responses Seibold et al., 2016). Based on findings from previous surveys con- (Chambers et al., 1999). Experimental harvest treatments included (1) ducted on the same snags (Chambers et al., 1997; Walter and Maguire, group selection, which represented localized, low intensity disturbance 2005), we also hypothesized that snags created in the clearcut and two- and resulted in 33% of the tree volume removed in 0.2 ha patches; (2) story treatments would receive more cumulative use by birds over time two-story, which represented evenly distributed moderate disturbance than those in the group selection treatment, as measured by the extent and resulted in 75% of the tree volume removed uniformly; and of cavity cover. We defined cavity cover as the cumulative area on the clearcut, which represented high intensity, stand-replacing disturbance snag bole that comprised of nesting, foraging, and natural cavities be- and resulted in all tree volume removed except for 1.3 live trees/ha. All cause (1) many snags in our study were extensively decayed, making it stands were replanted with Douglas-fir seedlings at a density of impossible to distinguish between cavity types, and (2) it is possible 625–865 trees/ha, depending on harvest treatment, and received her- that some cavities on created snags began as one cavity type (e.g., bicide applications 2–5 y after harvest to control competing vegetation nesting) but were expanded to become a different cavity type (e.g., (Chambers et al., 1997). foraging) during the lifetime of the snag. With respect to snag config- At the time stands were harvested, snags were created in either a uration, we hypothesized that snags created in clusters would experi- clustered configuration with 3–5 discrete groups within each stand, or ence greater use than scattered snags because snag proximity may in- scattered uniformly throughout each stand. In both configurations, crease foraging efficiency for birds. Under this assumption, we expected mean density of created snags was equal at the stand scale (3.8 snags/ snags to have greater external decay resulting from foraging activity ha). When available, natural snags were included in clusters of created and greater cavity cover within the clustered configuration, relative to snags (Chambers et al., 1997). Snags were created throughout each snags in the scattered configuration. Given the influence of foraging on stand except for the group selection treatment, where the small size of snag decay, we also predicted that snags created in clusters would have the harvested patches constrained snag creation to unharvested areas greater fall rates, break along the bole faster, and have more bark lost (Maguire and Chambers, 2005). All snags were created by topping live than scattered snags (Harmon et al., 1986; Jusino et al., 2016; Lorenz Douglas-fir trees with a chainsaw at a mean height of 17 m (minimum: et al., 2015). Alternatively, snags created in clusters may be more 15 m) and a mean diameter at breast height (DBH) of 75 cm (range: protected from wind than snags that are scattered and may therefore be 33–198 cm). Each snag was marked with a uniquely marked aluminum less likely to fall or break due to wind damage. As the first to quantify tag, allowing us to document changes in individual snags across time. the decay process of created snags in managed forests across ≥25 y, our Due to modification of treatments in some stands over time, only 26 of study provides a critical step in understanding the implications of a the 30 original stands were available for our study (group selection: widespread management technique for creating wildlife habitat that is n = 16, two-story: n = 5, clearcut: n = 5). commonly used within working forest landscapes.

146 A.M. Barry et al. Forest Ecology and Management 403 (2017) 145–151

2.3. Contemporary characteristics of created snags 2.5. Statistical methods

During January–March 2016, we revisited all individually marked We used a mixed linear modeling approach in the R (v3.3.1) sta- created Douglas-fir snags (n = 731) to assess their status (i.e., fallen or tistical environment to quantify treatment and time-specific variation in standing). From this total, 41 snags lacked historic measurements that snag characteristics. We used the ‘lme4’ package to construct general- were used as covariates in our model to assess influence of harvest ized linear mixed models with a binomial distribution and a logit link to treatment and configuration on snag characteristics (see below), re- examine the proportion of snags standing, the proportion of snags sulting in n = 690 snags available for analyses. We considered a snag to broken, and the proportion of snags with peeling bark. We constructed be standing if it was ≥2.5 m in height. For every snag we found linear mixed models to test for differences in contemporary estimates of standing, we also recorded whether it had broken along its bole. We mean bark cover and mean cavity cover among harvest treatments and classified a snag as broken by the appearance of an uneven top which between snag configurations. All models included harvest treatment (3 contrasted sharply with the smooth, even top that all snags had at the levels: group selection, two-story, clearcut), snag configuration (2 le- time of creation. Of the pool of snags that were still standing, we se- vels: clustered, scattered), a treatment × configuration interaction, and lected a random subset (n = 238) divided evenly among the three study block as fixed effects; stand as a random effect; and snag DBH and harvest treatments to quantify three additional characteristics by a ground slope as covariates. Both DBH and slope covariate data were single observer (AMB). For each snag, we estimated bark cover as the obtained at the time of snag creation and were included in models percent of the bole covered by bark (to the nearest 5%). We categorized because they are associated with Douglas-fir snag persistence (Huff and snags as having peeling bark (as a binary variable) when bark was Bailey, 2009). We log-transformed mean bark cover and mean cavity partially detached from ≥1m2 of the bole. To measure the cumulative cover to adhere with assumptions of normality and equal variance, and use of snags by cavity-excavating birds we estimated cavity cover as the back-transformed results for interpretation; we found no evidence of extent of all cavities observed on the bole (to the nearest 10%); we did overdispersion in our models. We used Tukey adjustments for all not include bark that was missing due to wood deterioration and lacked models with multiple comparisons, and we report least-squares mar- holes. We described a cavity as any excavation or depression in the snag ginal means for effect sizes and their associated 95% confidence in- that resulted from a nesting or foraging attempt or from a natural oc- tervals (CIs) with covariates set to their mean value. currence such as limb loss or fungal decay. Cavities may be used for To assess how snag characteristics changed over time, we compared multiple ecologically relevant functions over time (e.g., a nesting cavity historic data to contemporary data collected during the current study. may become a foraging area), so the origin of each cavity was difficult We compared the proportion of snags that had fallen, the proportion of to determine with certainty. Thus, we did not attempt to distinguish snags broken, and bark cover across all 4 time periods and we also between nesting, foraging, and natural cavities but instead combined compared contemporary estimates of the proportion of snags with them into a single estimate of cavity cover. This provided use with a peeling bark to estimates taken in 2001 using descriptive statistics. relative estimate of snag use under the assumption that a greater amount of cavity cover represented greater cumulative use of snags by 3. Results cavity-nesting birds. To calibrate cavity cover estimates, we first cal- culated cavity cover from digital photographs taken on a subset of 3.1. Contemporary characteristics of created snags created snags and used this to compare with visual estimates taken in the field on the same subset of snags before data collection began. We Across all treatments, 91% of all created snags were still standing then used photographs with cavity cover calculated in this manner as a and 65% of standing snags remained intact (i.e., were unbroken) ≥25 y visual reference during data collection to improve the accuracy of our after creation. The mean height of standing snags was 15.9 m (95% CI: estimates. We used two variables that were measured during surveys in 15.4, 16.5), indicating that standing snags lost very little of their tops 1990–1991 (i.e., snag DBH and ground slope) in our statistical models. since the time of creation. All snags that had fallen were found to have Snag DBH was measured at 1.4 m above ground height on all snags, and broken near their base, and no fallen snags showed evidence of having ground slope was measured as the slope of the ground averaged from been uprooted. We detected effects of harvest treatment on both the 20 m upslope and down-slope from each snag (Chambers et al., 1997). proportion of snags that were standing (X2 = 7.12, P = 0.03; Fig. 1a) and the proportion of snags that broke (X2 = 6.46, P = 0.04; Fig. 2a). The odds of a created snag remaining standing in the group selection 2.4. Long-term patterns of snag decay treatment were 2.7× greater ( ± 95% CI: 1.1, 6.4; z = 2.7, P = 0.02) than in the clearcut treatment. In contrast, we did not detect a differ- Initial characteristics of created snags were documented approxi- ence between the proportion of snags standing in the two-story treat- mately three months after each block was harvested in 1990–1991 ment compared with either the group selection (odds ratio [OR] (Chambers et al., 1997). Created snags were revisited at four points in = 1.7 ± 95% CI: 0.7, 4.1; z = 1.4, P = 0.34) or the clearcut treatment time (1995, 2001, 2008, and 2016) and data from those visits were used (OR = 1.6 ± 95% CI: 0.6, 4.3; z=1.1, P=0.52). The odds of a to provide a longitudinal assessment of snag characteristics (Chambers created snag being broken in the two-story treatment were 1.9× et al., 1997; Walter and Maguire, 2005; Huff and Bailey, 2009). To greater ( ± 95% CI: 1.1, 3.2; z = 2.9, P = 0.01) than in the group assess changes in snag characteristics over time, we compared con- selection treatment; however, there were similar proportions of snags temporary snag characteristics from our surveys (i.e., proportion broken when comparing between the clearcut treatment and either the standing, proportion with a broken bole, extent of bark cover, and ex- two-story (OR = 1.8 ± 95% CI: 0.9, 3.4; z = −2.0, P = 0.11) or the tent of bark peeling) to data from previous reports (i.e., Chambers et al., group selection treatments (OR = 0.9 ± 95% CI: 0.5, 1.6; z=0.34, 1997; Huff and Bailey, 2009; Walter and Maguire, 2005; hereafter, P=0.94). In contrast to an effect of harvest treatment, we did not historic data). We note that prior studies (Chambers et al., 1997; Walter detect an effect of snag configuration on either the proportion of snags and Maguire, 2005) reported the number of cavities observed on each standing (X2 = 0.02, P = 0.88; Fig. 1b) or on the proportion that broke snag, but this was not feasible during contemporary surveys because (X2 = 0.09, P = 0.77; Fig. 2b). We also note that we detected an effect decay in the intervening years did not allow for a strict delineation of of DBH on whether a snag was standing (X2 = 19.30, P < 0.001) or some cavities. In turn, this prevented us from making direct compar- had broken (X2 = 33.47, P < 0.001), but did not detect an effect of isons of the number of historic and contemporary cavities on created slope on either measure (X2 = 0.95, P = 0.33; X2 = 1.01, P = 0.32, snags. respectively). For the created snags for which we quantified additional

147 A.M. Barry et al. Forest Ecology and Management 403 (2017) 145–151

1.00 0.50 (a) A (a) B AB 0.45 A 0.95 0.40 AB 0.35 0.90 A 0.30

0.25 0.85

0.20 Mean proportion (± 95% CI) broken Mean proportion (± 95% Mean proportion (± 95% CI) standing Mean proportion (± 95%

0.80 0.15 Group selection Two-story Clearcut Group selection Two-story Clearcut

Harvest treatment Harvest treatment

1.00 0.45 (b) (b) A A A 0.40

0.95 0.35 A

0.90 0.30

0.25

0.85 0.20 Mean proportion (± 95% CI) broken Mean proportion (± 95% Mean proportion (± 95% CI) standing Mean proportion (± 95%

0.80 0.15 Clustered Scattered Clustered Scattered

Snag configuration Snag configuration

Fig. 1. Mean proportion ( ± 95% CI) of created snags that were standing within each (a) Fig. 2. Mean proportion ( ± 95% CI) of created snags that were broken within each (a) harvest treatment (group selection: n = 368, two-story: n = 165, clearcut: n = 157) and harvest treatment (group selection: n = 368, two-story: n = 165, clearcut: n = 157) and fi – (b) snag configuration (clustered: n = 398, scattered: n = 292) when measured 25–27 y (b) snag con guration (clustered: n = 398, scattered: n = 292) when measured 25 27 y after creation. Point estimates that do not share the same letter are considered to be after creation. Point estimates that do not share the same letter are considered to be ff statistically different from each other (P < 0.05). statistically di erent from each other (P < 0.05). characteristics (n = 238), we found that snags in treatments that ex- z = 4.86, P < 0.001) in the two-story treatment. Although all created perienced greater levels of harvest intensity and were subjected to an snags contained cavities, mean cavity cover was 1.4× greater in both open environment (i.e., two-story and clearcut) were generally more the clearcut ( ± 95% CI: 1.1, 1.9; t = 2.65; P = 0.03) and the two story similar to each other than to those created under lower harvest in- ( ± 95% CI: 1.0, 1.9; t = 2.46; P = 0.04) treatments relative to the tensity and where snags remained under a closed canopy (i.e., group group selection treatment. selection treatment; Table 1). We detected an effect of harvest treat- We did not detect an effect of snag configuration on bark cover (OR = 1.1 ± 95% CI: 0.9, 1.4; t = 1.67, P = 0.15) or cavity cover ment on bark cover (F2,18 = 28.35, P < 0.001), bark peeling 2 (OR = 1.1 ± 95% CI: 0.8, 1,4; t = 0.44, P = 0.68), but we did detect (X = 38.27, P < 0.001), and cavity cover (F2,18 = 10.30, P = 0.01). 2 Created snags in the group selection treatment typically exhibited less an effect on bark peeling (X = 5.90, P = 0.02). Snags in a clustered bark loss and bark peeling and had less cavity cover than snags in either configuration were 2.6 × more likely ( ± 95% CI: 1.2, 5.6; z = 2.54; the two-story or clearcut treatments. Mean bark cover in the group P = 0.01) to have bark peeling than in a scattered configuration. We 2 selection treatment was 1.4× greater ( ± 95% CI: 1.2, 1.6; t = 5.15; also detected an effect of DBH on bark cover (X = 8.36, P = 0.003), 2 2 P < 0.001) than in the clearcut treatment and 1.2× greater ( ± 95% bark peeling (X = 5.04, P = 0.03), and cavity cover (X = 24.47, CI: 1.0, 1.4; t = 2.93, P = 0.02) than in the two-story treatment. When P < 0.001). In contrast, we did not detect an effect of slope on bark 2 2 compared to the group selection treatment, the odds of a snag with cover (X = 0.02, P = 0.89), bark peeling (X = 0.07, P = 0.81), or 2 peeling bark were 14.5× greater ( ± 95% CI = 5.0, 41.8; z = 5.92, cavity cover (X = 0.17, P = 0.68). P < 0.001) in the clearcut and 9.0× greater ( ± 95% CI: 3.1, 25.9;

148 A.M. Barry et al. Forest Ecology and Management 403 (2017) 145–151

Table 1 Snag characteristics (i.e., mean bark cover, proportion of snags with peeling bark, mean cavity cover) measured on intentionally created snags relative to experimental harvest treatments 25–27 y after creation. See text for detailed definitions of snag characteristics.

Harvest treatment n Mean bark cover ( ± 95% CI) Snags with peeling bark ( ± 95% CI) Mean cavity cover ( ± 95% CI)

Group Selection 73 99.1% (86.5, 113.4) 16.7% (9.1, 28.7) 9.8% (7.7, 12.5) Two-story 69 81.1% (70.6, 93.4) 64.4% (51.0, 75.8) 13.7% (10.2, 18.3) Clearcut 96 71.1% (62.4, 81.0) 74.4% (63.7, 82.8) 13.9% (10.4, 18.5)

3.2. Long-term patterns of snag decay Harvest treatment also influenced cumulative use of snags over time by cavity-nesting birds, as indicated by the extent of cavity cover. Fall rate of created snags increased over time, with the greatest Previous surveys conducted on the same pool of snags in 1995 and 2001 increase between 2008 and 2016. No snags fell within the first 4–6y (Chambers et al., 1997; Walter and Maguire, 2005) were concordant to after creation, and only a single snag had fallen by 2001, 10–12 y after the findings from our study, with the highest estimates of bird-formed creation. By 2008, 99.5% of the snags were still standing although that cavities and more direct observations of nesting and foraging by cavity- decreased to 91% by 2016. Similarly, no standing snags had broken nesting birds in the two-story and clearcut treatments. These two within the first 4–6 y after creation, and only 1 had broken by 2001. treatments may have historically provided better habitat for some The percentage of standing snags that were broken was still low (3%) in woodpecker species, such as the Northern Flicker (Colaptes auratus), 2008, but increased markedly to 27% in 2016. Finally, the rate of bark that select open stands with lower live tree densities for nesting (Aitken loss also advanced with time since creation. The percent of snags with et al., 2002; Elchuk, et al., 2003; Walter and Maguire, 2005; Warren bark peeling away from the bole increased from 8% in 2001 to 54% in et al., 2005). Snags created in more open treatments may have also 2016. Although bark cover on snags remained high (97%) in 2001, it initially provided better foraging resources than snags in group selec- had decreased to 82% across treatments by 2016. The percentage of tion stands. Indeed, a recent study found that species richness of sa- snags that contained evidence of cavities increased from 88% in 2001 to proxylic beetles was higher in dead wood in open forest plots compared 100% in 2016. with closed forest plots (Seibold et al., 2016). If increases in wood- boring beetles did occur in more open treatments in our study, it could have attracted species that use snags as foraging substrates (e.g., 4. Discussion woodpeckers; Spring, 1965; Harmon et al., 1986) and led to greater areas of cavity creation and use. Foraging and nesting by woodpeckers 4.1. Contemporary characteristics of created snags may have further contributed to the contemporary differences in snag characteristics we observed by changing the structure of snags as they Our study found that the great majority of Douglas-fir snags created drilled through bark and sapwood to obtain food and create nesting by topping remained standing ≥25 y after they were created. The size cavities (Bull et al., 1983). In the process of excavating, woodpeckers of the snags in our study (mean DBH = 75 cm) likely enhanced their can directly expose heartwood to fungi and other decay organisms persistence because large diameter natural Douglas-fir snags can remain (Jusino et al., 2016), thereby accelerating physical and biological snag standing for > 100 y in the Oregon Coast Range (Cline et al., 1980). In decay in the historically more open treatments. Although we were addition, Douglas-fir snags may remain standing longer than other tree unable to distinguish between the effects of the microclimate and bird species due in part to their higher ratio of heartwood to sapwood, which use on snag decay, this should be an important focus of other studies as has higher resistance to fungi and thus greater resistance to decay they are both likely to contribute to the patterns we found. (Kimmey and Furniss, 1943; Harmon et al., 1986; Hallett et al., 2001). The minimal effect that we observed of spatial configuration on Furthermore, all created snags had indications of use by birds over nearly all snag characteristics measured is consistent with previous time, perhaps because topped snags receive more use by birds than work (Chambers et al., 1997; Walter and Maguire, 2005; Arnett et al., snags created through within 4–7 y since creation (Hallett 2010; Hane et al., 2012); the lone exception was the proportion of snags et al., 2001). Taken together, this suggests that both tree species and that contained peeling bark. Clustered snags may be more likely to method of creation are important considerations for determining attract beetles or decomposer organisms that accelerate decay and longevity and usefulness of created snags, and that topped Douglas-fir cause sapwood to break down more quickly (Chamberlin, 1918; Angers trees serve as useful habitat features for cavity-nesting birds within et al., 2012; Seibold et al., 2016), and the activity of these organisms managed forests. can often lead to a greater propensity for bark to peel away from cre- Although the proportion of created snags that were still standing ated snags. An increase in insect density on clustered snags may also during the course of our study was high, we did find that silvicultural attract more foraging birds, such as woodpeckers, that drill for prey practices (e.g., harvest treatment at time of creation) had a strong in- (Raphael and White, 1984), resulting in a similar effect on snag decay fluence on the propensity to break and current-day bark characteristics. and subsequent bark integrity. However, we did not detect any differ- The greater levels of bark loss in treatments with fewer live trees re- ences in cavity cover or bark cover between snag configurations, so the tained at harvest may have resulted from changes in microclimatic differences in bark integrity we detected are likely due to some other conditions that were induced by a reduction in canopy cover and sur- factor(s). To the best of our knowledge, there are no similar studies that rounding tree densities, which are typically more variable in open have examined the effect of spatial configuration on characteristics of stands (Harmon et al., 1986; Garber et al., 2005. For example, snags in created snags, so this topic warrants further exploration. the two-story and clearcut treatments were less protected and subjected to more extreme fluctuations in moisture and temperature than those in the group selection treatments. This may have led to greater decay rates 4.2. Long-term patterns of snag decay via enhanced stress to wood cells, in turn increasing breaking and bark detachment from the bole of the snag (Harmon et al., 1986). In contrast, When historic data are considered (Chambers et al., 1997; Walter snags in the group selection treatment have experienced a more stable and Maguire, 2005; Huff and Bailey, 2009), it is evident that the rate of environment with dampened fluctuations in moisture and temperature breaking and bark loss on snags across all harvest treatments was in- within the matrix of mature forest canopy, ultimately leading to lower itially slow and increased markedly in the second decade since creation. rates of decay we observed in our study. Between 17–19 y (Huff and Bailey, 2009) and 25–27 y (this study) after

149 A.M. Barry et al. Forest Ecology and Management 403 (2017) 145–151 snags were created, breaking and bark peeling had increased sub- Arnett, E.B., Kroll, A.J., Duke, S.D., 2010. Avian foraging and nesting use of created snags in intensively-managed forests of western Oregon, USA. For. Ecol. Manage. 260, stantially across all harvest treatments. Breaking and bark loss are ex- 1773–1779. ternal characteristics that have been commonly used to describe more Barry, A.M., 2017. Created snag dynamics and influence on cavity-nesting bird com- advanced stages of snag decay, and our findings match other studies of munties over 25 years in Western Oregon. Masters thesis Oregon State University. fi Bull, E.L., Partridge, A.D., 1986. Methods of killing trees for use by cavity nesters. Wildl. natural Douglas- r snags in our region which have reported extensive Soc. Bull. 14, 142–146. breaking and bark loss when snags were between 19 and 50 y old (Cline Bunnell, F.L., 2013. Sustaining cavity-using species: patterns of cavity use and implica- et al., 1980). Whereas breaking and bark loss may be important factors tions to forest management. ISRN For. 1–33 Article ID 457698. fi in the availability and selection of habitat by wildlife (Bunnell, 2013; Chamberlin, W., 1918. Bark-beetles infesting the Douglas r. Oregon Agric. Exp. Station Bull. 147, 1–40. Chambers et al., 2002; Franklin et al., 1987; Geleynse et al., 2016; Chambers, C., Carrigan, T., Sabin, T., Tappeiner, J., McComb, W., 1997. Use of artificially Harmon et al., 1986; Rose et al., 2001), external characteristics of snags created Douglas-fir snags by cavity-nesting birds. West. J. Appl. For. 12, 93–97. have been shown to be poorly correlated with internal decay (Schepps Chambers, C.L., 1996. Response of terrestrial vertebrates to three silvicultural treatments in the Central Oregon Coast Range. Ph.D. thesis Oregon State University. et al., 1999). Thus, future studies should seek to combine measurements Chambers, C.L., McComb, W.C., Tappeiner, J.C., 1999. Breeding bird responses to three of internal wood hardness with external characteristic of created snags sylvicultural treatments in the Oregon coast range. Ecol. Appl. 9, 171–185. to better understand how external characteristics relate to snag decay, Chambers, C.L., Alm, V., Sliders, M.S., Rabe, M., 2002. Use of artificial roosts by forest- dwelling bats in Northern Arizona. Wildl. Soc. Bull. 30 (2002), 1085–1091. which was not possible in our study because of safety considerations. Cline, S.P., Berg, A.B., Wight, H.M., 1980. Snag characteristics and dynamics in Douglas- Although we were unable to assess changes in cavity cover over time, Fir forests, Western Oregon. J. Wildlife Manage. 44, 773–786. we note that current-day estimates of cavity cover did not exceed 60% Duane, M., 2001. Response of wood-boring beetles (Coleoptera: Buprestidae, Cerambycidae) to prescribed understory burning in mixed-conifer stands in (Barry, 2017), suggesting that snags may not remain standing above Southwestern Oregon. Masters thesis Oregon State University. this threshold, perhaps due to structural instability caused by removing Elchuk, C.L., Wiebe, K.L., Wiebe, E., 2003. Home-range size of Northern Flickers (Colaptes structural components of snags. auratus) in relation to habitat and parental attributes. Can. J. Zool. 81, 954–961. Franklin, J.F., Shugart, H.H., Harmon, M.E., 1987. Tree death as an ecological process: the causes, consequences, and variability of tree mortality. Bioscience 37, 550–556. 4.3. Conclusions and management implications Garber, S.M., Brown, J.P., Wilson, D.S., Maguire, D.A., Heath, L.S., 2005. Snag longevity under alternative silvicultural regimes in mixed-species forests of central Maine. Can. – Although most of the snags we examined were still standing ≥25 y J. For. Res. 35, 787 796. Geleynse, D.M., Nol, E., Burke, D.M., Elliott, K.A., 2016. Brown Creeper (Certhia amer- after creation, snags had an increased rate of breaking and experienced icana) demographic response to hardwood forests managed under the selection changes in characteristics that appear to be associated with advanced system. Can. J. For. Res. 46, 499–507. ’ decay. Because snags created in treatments with greater harvest in- Hallett, J.G., Lopez, T., O Connell, M.A., Borysewicz, M.A., 2001. Decay dynamics and avian use of artificially created snags. Northwest Sci. 75, 378–386. tensity exhibited greater likelihood of falling and more bark loss, forest Hane, M.E., Kroll, A.J., Johnson, J.R., Rochelle, M., Arnett, E.B., 2012. Experimental managers whose goal is wildlife habitat creation may consider leaving effects of structural enrichment on avian nest survival. For. Ecol. Manage. 282, some live trees at the time of snag creation to provide long-term habitat 167–174. Harmon, M.E., Franklin, J.F., Swanson, F.J., Sollins, P., Gregory, S.V., Lattin, J.D., et al., features for snag-dependent species. If snags created under a mature 1986. Ecology of course woody debris in temperate ecosystems. Adv. Ecol. Res. 15, forest matrix have greater persistence and experience lower rates of 1–436. decay than snags created in the open, as was the case in our study, then Huff, T.D., Bailey, J.D., 2009. Longevity and dynamics of fatally and nonfatally topped Douglas-fir in the Coast Range of Oregon. Can. J. For. Res. 39, 2224–2233. they may represent important structures that can be used by wildlife Jusino, M.A., Lindner, D.L., Banik, M.T., Rose, K.R., Walters, J.R., 2016. Experimental across long timescales. A critical but unanswered question is the extent evidence of a symbiosis between red-cockaded woodpeckers and fungi. Proc. R. Soc. to which created snags are used by cavity-nesting birds for foraging and B: Biol. Sci. 283, 20160106. ≥ Kilgo, J.C., Vukovich, M.A., 2014. Can snag creation benefit a primary cavity nester: nesting after 25 y. The increased rate of breaking and bark loss that response to an experimental pulse in snag abundance. Biol. Cons. 171, 21–28. we observed in the open treatments suggests that the usefulness of these Kimmey, J.W., Furniss, R.L., 1943. Deterioration of Fire-killed Douglas-fir. USDA Tech. snags for some cavity-nesting birds, such as woodpeckers, may decline Bull, Washington, D.C. through time. Nevertheless, some species use snags in later stages of Kluber, M., 2007. Terrestrial amphibian distribution, habitat associations and downed wood temperature profiles in managed headwater forests with riparian buffers in the decay, so they may benefit from snags containing greater decay in Oregon Coast Range. Masters thesis Oregon State University. managed stands that result from a combination of different harvest Kroll, A.J., Lacki, M.J., Arnett, E.B., 2012. Research needs to support management and fi treatments. conservation of cavity-dependent birds and bats on forested landscapes in the Paci c Northwest. West. J. Appl. For. 27, 128–136. Lorenz, T.J., Vierling, K.T., Johnson, T.R., Fischer, P.C., 2015. The role of wood hardness Acknowledgements in limiting nest site selection in avian cavity excavators. Ecol. Appl. 25, 1016–1033. Maguire, C., Chambers, C., 2005. College of Forestry Integrated Research Project: Ecological and Socioeconomic Responses to Alternative Silvicultural Treatments. Financial support for this work was provided by the Fish and Oregon State University, Corvallis. Wildlife Habitat in Managed Forests Research Program in the College of Manning, J.A., Edge, W.D., 2004. Small mammal survival and downed wood at multiple Forestry at Oregon State University, the US Geological Survey Forest scales in managed forests. J. Mammal. 85, 87–96. McComb, B.W.C., Spies, T.A., Emmingham, W.H., 1993. Douglas-fir forests: managing for and Rangeland Ecosystem Science Center, and the Oregon State timber and mature-forest habitat. J. For. 91, 31–42. University College Forests. We thank S. Walter for sharing unpublished Morrison, M.L., Raphael, M.G., 1993. Modeling the dynamics of snags. Ecol. Appl. 3, data; J. Kiser, M. Harmon, B. McComb, S. Fitzgerald, and B. Klumph for 322–330. Newton, I., 1994. The role of nest sites in limiting the numbers of hole-nesting birds: a logistical support; A. Muldoon and L. Ganio for statistical advice; N. review. Biol. Cons. 70, 265–276. Garlick, C. Mathis, K. Garcia, M. Wagner, J. Hollenbeck and J. Johnson Raphael, M.G., White, M., 1984. Use of snags by cavity-nesting birds in the Sierra Nevada. for assistance in the field; and C. Chambers and two anonymous re- Wildlife Monogr. 86, 3–66. viewers for helpful feedback on a previous version of this manuscript. Rose, C.L., Marcot, B.G., Mellen, T.K., Ohmann, J.L., Waddell, K.L., Lindley, D.L., Schreiber, B., 2001. Decaying Wood in Paci fic Northwest Forests: Concepts and Tools Any use of trade, firm, or product names is for descriptive purposes only for Habitat Management. Wildlife-habitat Relationships in Oregon and Washington. and does not imply endorsement by the U.S. Government. Oregon State University Press, Corvallis. Schepps, J., Lohr, S., Martin, T.E., 1999. Does tree hardness influence nest-tree selection by primary cavity nesters? Auk 166 (3), 658–665. References Schreiber, B., DeCalesta, D., 1992. The relationship between cavity-nesting birds and snags on clearcuts in western Oregon. For. Ecol. Manage. 50, 299–316. Aitken, K.E.H., Aitken, K.E.H., Wiebe, K.L., Wiebe, K.L., Martin, K., Martin, K., 2002. Seibold, S., Bassler, C., Brandl, R., Buche, B., Szallies, A., Thorn, S., et al., 2016. Nest-site reuse patterns for a cavity-nesting bird community in interior British Microclimate and habitat heterogeneity as the major drivers of beetle diversity in – Columbia. Auk 119, 391–402. dead wood. J. Appl. Ecol. 53, 934 943. Angers, V.A., Drapeau, P., Bergeron, Y., 2012. Mineralization rates and factors influen- Spring, B.L.W., 1965. Climbing and pecking adaptations in some North American – cing snag decay in four North American boreal tree species. Can. J. For. Res. 42, woodpeckers. Condor 67, 457 488. 157–166. Swanson, F.J., Franklin, J.F., 1992. New forestry principles from ecosystem analysis of

150 A.M. Barry et al. Forest Ecology and Management 403 (2017) 145–151

Pacific Northwest forests. Ecol. Appl. 2, 262–274. in western Oregon. J. Wildlife Manage. 69, 1578–1591. Thomas, J.W., 1979. Wildlife Habitats in Managed Forests: the Blue Mountains of Oregon Warren, T.L., Betts, M.G., Diamond, A.W., Forbes, G.J., 2005. The influence of local ha- and Washington, vol. 553. U.S. Department of Agriculture. bitat and landscape composition on cavity-nesting birds in a forested mosaic. For. Waldien, D.L., Hayes, J.P., Huso, M.M.P., 2006. Use of downed wood by Townsend's Ecol. Manage. 214, 331–343. chipmunks (Tamias townsendii) in Western Oregon. J. Mammal. 87, 454–460. Wilhere, G.F., 2003. Simulations of snag dynamics in an industrial Douglas-fir forest. For. Walter, S.T., Maguire, C.C., 2005. Snags, cavity-nesting birds, and silvicultural treatments Ecol. Manage. 174, 521–539.

151